On yet another hand, most people have powerful inhibitions against killing, so that might be harder to train than protecting children.

It is very hard to teach people to kill -- a fact that's largely responsible for the lopsided casualty ratios in engagements between well-trained and poorly trained armies. Most of the research that has been done on this is in the context of police and military training, though, where the main motivating factor is not letting your buddies down; I'd expect the psychology to be somewhat different in the context of defending children. Given that teachers self-select for very different psychology than cops or soldiers, though, I'm not sure which way the statistics would end up running.